home / skills / openclaw / skills / infinite-memory-v2

infinite-memory-v2 skill

safe

This skill provides high-precision memory with long-term recall to retrieve and store contextual data across extensive conversations.

npx playbooks add skill openclaw/skills --skill infinite-memory-v2

Review the files below or copy the command above to add this skill to your agents.

Files (9)

SKILL.md

824 B

---
name: infinite-memory
version: 1.0.0
description: High-precision memory with 100% recall accuracy for long contexts.
emoji: 🦞
metadata:
  clawdbot:
    requires:
      bins: 
        - python
        - curl
    files:
      - scripts/recall.py
      - scripts/ingest.py
      - engine/memory_engine_parallel_lms.py
      - references/AUTO_INTEGRATION.md
      - memory_service.py
      - requirements.txt
---

# Infinite Memory 🦞

High-precision RAG engine for deep context retrieval (Phase 16 Architecture).

## Tools

### recall_facts
- **Cmd:** `python scripts/recall.py "{{query}}"`
- **Goal:** Search for facts in the historical database.

### memorize_data
- **Cmd:** `python scripts/ingest.py "{{filename}}" "{{text}}"`
- **Goal:** Store new data into the long-term memory.

Overview

This skill provides a high-precision long-term memory engine with 100% recall accuracy for very large contexts. It archives and retrieves historical facts and versions from a centralized archive of skills, enabling precise retrieval for deep-context workflows. The system is optimized for deterministic recall and durable storage of ingestible items.

How this skill works

The skill exposes two core operations: recall_facts to query the archived database for exact facts and memorize_data to ingest and persist new items into long-term memory. Queries run against a versioned archive so results include historical context and provenance. Ingested items are indexed and stored to support reliable retrieval across arbitrarily long conversations.

When to use it

You need deterministic, 100% accurate recall of prior facts or archived skill versions.
Working with very long contexts where in-session memory is insufficient.
Auditing, debugging, or tracing historical changes across skill versions.
Backing up important facts, documents, or conversation checkpoints for later retrieval.

Best practices

Provide concise, unique keys or queries when ingesting so retrieval is unambiguous.
Batch related items together to preserve context and reduce lookup complexity.
Include provenance metadata (source, timestamp, version) with every ingest.
Use precise query phrases when calling recall_facts to avoid broad matches.
Validate retrieved facts against expected provenance before acting on them.

Example use cases

Restore a precise configuration or snippet from a past skill version during debugging.
Store critical decisions or meeting notes and recall them exactly weeks or months later.
Build a RAG pipeline that requires exact historical citations and provenance.
Back up user preferences or legal statements that must be retrieved verbatim.

FAQ

What operations are available?

Two operations: recall_facts for queries and memorize_data for ingestion.

How is 100% recall accuracy ensured?

Data is indexed and versioned in a deterministic archive so exact stored items are returned for matching queries.